Bright Challenge Advances All-Weather Building Damage Mapping with SAR and Optical Imagery
The Bright Challenge enhances all-weather mapping of building damage by integrating submeter-resolution optical images taken before events with post-event synthetic aperture radar (SAR) data. This initiative expanded the Bright dataset to include instance-level annotations for about 291,000 structures across 16 different disaster events and seven types of disasters. Competitors were tasked with identifying and outlining each building, assigning one of three exclusive damage classifications: intact, damaged, or destroyed. The concluding phase focused solely on two 2025 events not included in the training: a California wildfire and a hurricane in Jamaica. In total, 157 participants submitted 1,289 entries, with 46 teams qualifying for the final stage, addressing the urgent need for rapid post-disaster assessments when optical images are hindered by clouds, smoke, or darkness.
Key facts
- The Bright Challenge evaluates all-weather building damage mapping using pre-event optical and post-event SAR imagery.
- The dataset includes instance-level annotations for about 291,000 buildings across 16 disaster events.
- Seven disaster types are represented in the dataset.
- Participants must assign one of three damage labels: intact, damaged, or destroyed.
- The final phase tested on a 2025 California wildfire and a 2025 Jamaica hurricane.
- 157 participants made 1,289 submissions; 46 teams entered the final phase.
- The challenge addresses the problem of unavailable optical imagery due to cloud, smoke, or darkness.
- The work is published on arXiv under identifier 2607.22746.
Entities
Institutions
- arXiv
- Bright
Locations
- California
- United States
- Jamaica